1

Text Categorization

using Naïve Bayes

Mausam

(based on slides of Dan Weld, Dan

Jurafsky, Prabhakar Raghavan, Hinrich

Schutze, Guillaume Obozinski, David D.

Lewis, Fei Xia)

2

Categorization

• Given:

– A description of an instance, xX, where X is

the instance language or instance space.

– A fixed set of categories:

C={c1, c2,…cn}

• Determine:

– The category of x: c(x)C, where c(x) is a

categorization function whose domain is X and

whose range is C.

County vs. Country?

3

Who wrote which Federalist papers?

• 1787-8: anonymous essays try to convince

New York to ratify U.S Constitution: Jay,

Madison, Hamilton.

• Authorship of 12 of the letters in dispute

• 1963: solved by Mosteller and Wallace

using Bayesian methods

James Madison Alexander Hamilton

Male or female author?• The main aim of this article is to propose an exercise in stylistic analysis

which can be employed in the teaching of English language. It details the

design and results of a workshop activity on narrative carried out with

undergraduates in a university department of English. The methods proposed

are intended to enable students to obtain insights into aspects of cohesion and

narrative structure: insights, it is suggested, which are not as readily obtainable

through more traditional techniques of stylistic analysis.

• My aim in this article is to show that given a relevance theoretic approach to

utterance interpretation, it is possible to develop a better understanding of

what some of these so-called apposition markers indicate. It will be argued

that the decision to put something in other words is essentially a decision

about style, a point which is, perhaps, anticipated by Burton-Roberts when he

describes loose apposition as a rhetorical device. However, he does not justify

this suggestion by giving the criteria for classifying a mode of expression as a

rhetorical device.S. Argamon, M. Koppel, J. Fine, A. R. Shimoni, 2003. “Gender, Genre, and Writing Style in Formal Written Texts,” Text, volume 23, number 3,

pp. 321–346

Female writers use

more first person/second person pronouns

more gender laiden third person pronouns

(overall more personalization)

Positive or negative movie review?

• unbelievably disappointing

• Full of zany characters and richly applied

satire, and some great plot twists

• this is the greatest screwball comedy ever

filmed

• It was pathetic. The worst part about it was

the boxing scenes.

•6

What is the subject of this article?

• Antogonists and

Inhibitors

• Blood Supply

• Chemistry

• Drug Therapy

• Embryology

• Epidemiology

• …•7

MeSH Subject Category Hierarchy

?

MEDLINE Article

Text Classification

• Assigning documents to a fixed set of categories, e.g.

• Web pages

– Yahoo-like classification

– Assigning subject categories, topics, or genres

• Email messages

– Spam filtering

– Prioritizing

– Folderizing

• Blogs/Letters/Books

– Authorship identification

– Age/gender identification

• Reviews/Social media

– Language Identification

– Sentiment analysis

– …

Classification Methods:

Hand-coded rules

• Rules based on combinations of words or other features

– spam: black-list-address OR (“dollars” AND “have been selected”)

• Accuracy can be high

– If rules carefully refined by expert

• But building and maintaining these rules is expensive

10

Learning for Text Categorization

• Hard to construct text categorization functions.

• Learning Algorithms:

– Bayesian (naïve)

– Neural network

– Logistic Regression

– Nearest Neighbor (case based)

– Support Vector Machines (SVM)

18

Bayesian Methods

• Learning and classification methods based on probability theory.

– Bayes theorem plays a critical role in probabilistic learning and classification.

– Uses prior probability of each category given no information about an item.

• Categorization produces a posteriorprobability distribution over the possible categories given a description of an item.

Bayes’ Rule Applied to Documents and

Classes

P(c | d) =P(d | c)P(c)

P(d)

• For a document d and a class c

Naïve Bayes Classifier (I)

cMAP = argmaxcÎC

P(c | d)

= argmaxcÎC

P(d | c)P(c)

P(d)

= argmaxcÎC

P(d | c)P(c)

MAP is “maximum a

posteriori” = most likely

class

Bayes Rule

Dropping the

denominator

Naïve Bayes Classifier (II)

cMAP = argmaxcÎC

P(d | c)P(c)

Document d

represented as

features x1..xn)()|,,,(argmax 21 cPcxxxP nCc

Naïve Bayes Classifier (IV)

How often does this class

occur?

)()|,,,(argmax 21 cPcxxxPc nCc

MAP

O(|X|n•|C|) parameters

We can just count the

relative frequencies in a

corpus

Could only be estimated if a very, very large number of training examples was available.

Multinomial Naïve Bayes Independence

Assumptions

)|,,,( 21 cxxxP n

• Bag of Words assumption: Assume position doesn’t matter

The bag of words representation

I love this movie! It's sweet, but with satirical humor. The dialogue is great and the adventure scenes are fun… It manages to be whimsical and romantic while laughing at the conventions of the fairy tale genre. I would recommend it to just about anyone. I've seen it several times, and I'm always happy to see it again whenever I have a friend who hasn't seen it yet.

γ( )=c

I love this movie! It's sweet, but with satirical humor. The dialogue is great and the adventure scenes are fun… It manages to be whimsical and romantic while laughing at the conventions of the fairy tale genre. I would recommend it to just about anyone. I've seen it several times, and I'm always happy to see it againwhenever I have a friend who hasn't seen it yet.

γ( )=c

The bag of words representation

The bag of words representation:

using a subset of words

x love xxxxxxxxxxxxxxxx sweetxxxxxxx satirical xxxxxxxxxxxxxxxxxxxxx great xxxxxxxxxxxxxxxxxxxxxxxxxx fun xxxxxxxxxxxxxxxxx whimsical xxxxromantic xxxx laughingxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx recommend xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx several xxxxxxxxxxxxxxxxxxxxxx happy xxxxxxxxx againxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

γ( )=c

γ( )=cgreat 2

love 2

recommend 1

laugh 1

happy 1

... ...

The bag of words representation

Planning GUIGarbageCollection

Machine

Learning NLP

parser

tag

training

translation

language...

learning

training

algorithm

shrinkage

network...

garbage

collection

memory

optimization

region...

Test

document

parser

language

label

translation

…

Bag of words for document

classification

...planning

temporal

reasoning

plan

language...

?

Multinomial Naïve Bayes Classifier

)()|,,,(argmax 21 cPcxxxPc nCc

MAP

cNB = argmaxcÎC

P(c j ) P(x | c)xÎX

Õ

Multinomial Naïve Bayes Independence

Assumptions

)|,,,( 21 cxxxP n

• Bag of Words assumption: Assume position doesn’t matter

• Conditional Independence: Assume the feature probabilities P(xi|cj) are independent given the class c.

)|(...)|()|()|()|,,( 3211 cxPcxPcxPcxPcxxP nn

Applying Multinomial Naive Bayes

Classifiers to Text Classification

cNB = argmaxcjÎC

P(c j ) P(xi | c j )iÎpositions

Õ

positions all word positions in test document

Learning the Multinomial Naïve Bayes Model

• First attempt: maximum likelihood estimates

– simply use the frequencies in the data

P̂(wi | c j ) =count(wi,c j )

count(w,c j )

wÎV

å

P̂(c j ) =doccount(C = c j )

Ndoc

• Create mega-document for topic j by concatenating all docs in this topic– Use frequency of w in mega-document

Parameter estimation

fraction of times word wi appears among all words in documents of topic cj

P̂(wi | c j ) =count(wi,c j )

count(w,c j )

wÎV

å

Problem with Maximum Likelihood

• What if we have seen no training documents with the word fantastic and classified in the topic positive (thumbs-up)?

• Zero probabilities cannot be conditioned away, no matter the other evidence!

P̂("fantastic" positive) = count("fantastic", positive)

count(w, positive

wÎV

å ) = 0

cMAP = argmaxc P̂(c) P̂(xi | c)i

Õ

Sec.13.3

Laplace (add-1) smoothing for Naïve Bayes

P̂(wi | c) =count(wi,c)+1

count(w,c)+1( )wÎV

å

=count(wi,c)+1

count(w,c

wÎV

å )æ

èçç

ö

ø÷÷ + V

P̂(wi | c) =count(wi,c)

count(w,c)( )wÎV

å

Multinomial Naïve Bayes: Learning

• Calculate P(cj) terms– For each cj in C do

docsj all docs with class =cj

P(wk | c j ) ¬nk +a

n+a |Vocabulary |

P(c j ) ¬| docs j |

| total # documents|

• Calculate P(wk | cj) terms• Textj single doc containing all docsj

• For each word wk in Vocabularynk # of occurrences of wk in Textj

• From training corpus, extract Vocabulary

47

Naïve Bayes Time Complexity

• Training Time: O(|D|Ld + |C||V|)) where Ld is the average length of a document in D.

– Assumes V and all Di , ni, and nij pre-computed in O(|D|Ld) time during one pass through all of the data.

– Generally just O(|D|Ld) since usually |C||V| < |D|Ld

• Test Time: O(|C| Lt) where Lt is the average length of a test document.

• Very efficient overall, linearly proportional to the time needed to just read in all the data.

48

Easy to Implement

• But…

• If you do… it probably won’t work…

Probabilities: Important Detail!

We are multiplying lots of small numbers

Danger of underflow!

0.557 = 7 E -18

Solution? Use logs and add!

p1 * p2 = e log(p1)+log(p2)

Always keep in log form

49

Generative Model for Multinomial Naïve Bayes

•50

c=China

X1=Shanghai X2=and X3=Shenzhen X4=issue X5=bonds

Naïve Bayes and Language Modeling

• Naïve bayes classifiers can use any sort of feature

– URL, email address, dictionaries, network features

• But if, as in the previous slides

– We use only word features

– we use all of the words in the text (not a subset)

• Then

– Naïve bayes has an important similarity to language

modeling.

•51

Each class = a unigram language model

• Assigning each word: P(word | c)

• Assigning each sentence: P(s|c)=Π P(word|c)

0.1 I

0.1 love

0.01 this

0.05 fun

0.1 film

…

I love this fun film

0.1 0.1 .05 0.01 0.1

Class pos

P(s | pos) = 0.0000005

Sec.13.2.1

Naïve Bayes as a Language Model

• Which class assigns the higher probability to s?

0.1 I

0.1 love

0.01 this

0.05 fun

0.1 film

Model pos Model neg

filmlove this funI

0.10.1 0.01 0.050.10.10.001 0.01 0.0050.2

P(s|pos) > P(s|neg)

0.2 I

0.001 love

0.01 this

0.005 fun

0.1 film

Sec.13.2.1

Example: Naïve Bayes in Spam Filtering

• SpamAssassin Features:– Mentions Generic Viagra

– Online Pharmacy

– Mentions millions of (dollar) ((dollar) NN,NNN,NNN.NN)

– Phrase: impress ... girl

– From: starts with many numbers

– Subject is all capitals

– HTML has a low ratio of text to image area

– One hundred percent guaranteed

– Claims you can be removed from the list

– 'Prestigious Non-Accredited Universities'

– http://spamassassin.apache.org/tests_3_3_x.html

Advantages

• Simple to implement

– No numerical optimization, matrix algebra, etc

• Efficient to train and use

– Easy to update with new data

– Fast to apply

• Binary/multi-class

• Good in domains with many equally important features

– Decision Trees suffer from fragmentation in such cases –

especially if little data

• Comparatively good effectiveness with small training sets

• A good dependable baseline for text classification

– But we will see other classifiers that give better accuracy

55

Disadvantages

• Independence assumption wrong

– Absurd estimates of class probabilities

• Output probabilities close to 0 or 1

– Thresholds must be tuned; not set analytically

• Generative model

– Generally lower effectiveness than

discriminative techniques

56

Experimental Evaluation

Question: How do we estimate the

performance of classifier on unseen data?

• Can’t just at accuracy on training data – this

will yield an over optimistic estimate of

performance

• Solution: Cross-validation

• Note: this is sometimes called estimating

how well the classifier will generalize

57

Evaluation: Cross Validation

• Partition examples into k disjoint sets

• Now create k training sets

– Each set is union of all equiv classes except one

– So each set has (k-1)/k of the original training data Train

Test

Test

Test

…

58

Cross-Validation (2)

• Leave-one-out

– Use if < 100 examples (rough estimate)

– Hold out one example, train on remaining examples

• 10-fold

– If have 100-1000’s of examples

• M of N fold

– Repeat M times

– Divide data into N folds, do N fold cross-validation

59

Evaluation Metrics

• Accuracy: no. of questions correctly answered

• Precision (for one label): accuracy when classification = label

• Recall (for one label): measures how many instances of a label were

missed.

• F-measure (for one label): harmonic mean of precision & recall.

• Area under Precision-recall curve (for one label): vary parameter to

show different points on p-r curve; take the area

60

Precision & Recall

P

N

“P” “N”

TP FN

FP TN

Predicted

Act

ual

Two class situation

FP

FPTP

Multi-class situation:

Precision = TP/(TP+FP)

Recall = TP/(TP+FN)

F-measure = 2pr/(p+r)

61

A typical precision-recall curve

62

Construct Better Features

• Key to machine learning is having good

features

• In gen 2 ML, large effort devoted to

constructing appropriate features

• Ideas??

63

Issues in document representation

• Cooper’s vs. Cooper vs. Coopers.

• Full-text vs. full text vs. {full, text} vs. fulltext.

• résumé vs. resume.

Cooper’s concordance of Wordsworth was published in

1911. The applications of full-text retrieval are legion:

they include résumé scanning, litigation support and

searching published journals on-line.

slide from Raghavan, Schütze,

Larson

Punctuation

• Ne’er: use language-specific, handcrafted

“locale” to normalize.

• State-of-the-art: break up hyphenated

sequence.

• U.S.A. vs. USA

• a.out

slide from Raghavan, Schütze,

Larson

Numbers

• 3/12/91

• Mar. 12, 1991

• 55 B.C.

• B-52

• 100.2.86.144

– Generally, don’t represent as text

– Creation dates for docs

slide from Raghavan, Schütze,

Larson

Possible Feature Ideas

• Look at capitalization (may indicated a

proper noun)

• Look for commonly occurring sequences

• E.g. New York, New York City

• Limit to 2-3 consecutive words

• Keep all that meet minimum threshold (e.g.

occur at least 5 or 10 times in corpus)

67

Case folding

• Reduce all letters to lower case

• Exception: upper case in mid-sentence

– e.g., General Motors

– Fed vs. fed

– SAIL vs. sail

slide from Raghavan, Schütze,

Larson

Thesauri and Soundex

• Handle synonyms and spelling variations

– Hand-constructed equivalence classes

• e.g., car = automobile

slide from Raghavan, Schütze,

Larson

Spell Correction

• Look for all words within (say) edit distance

3 (Insert/Delete/Replace) at query time

– e.g., arfiticial inteligence

• Spell correction is expensive and slows the

processing significantly

– Invoke only when index returns zero matches?

slide from Raghavan, Schütze,

Larson

Lemmatization

• Reduce inflectional/variant forms to base form

– am, are, is be

– car, cars, car's, cars' car

the boy's cars are different colors

the boy car be different color

72

Stemming

• Are there different index terms?

– retrieve, retrieving, retrieval, retrieved, retrieves…

• Stemming algorithm:

– (retrieve, retrieving, retrieval, retrieved, retrieves)

retriev

– Strips prefixes of suffixes (-s, -ed, -ly, -ness)

– Morphological stemming

• Problems: sand / sander & wand / wander

Copyright © Weld 2002-2007 72

73

Stemming Continued

• Can reduce vocabulary by ~ 1/3

• C, Java, Perl versions, python, c#www.tartarus.org/~martin/PorterStemmer

• Criterion for removing a suffix – Does "a document is about w1" mean the same as

– a "a document about w2"

• Problems: sand / sander & wand / wander

• Commercial SEs use giant in-memory tables

Copyright © Weld 2002-2007 73

•74

Features

• Domain-specific features and weights: very important in real performance

• Upweighting: Counting a word as if it occurred twice:

– title words (Cohen & Singer 1996)

– first sentence of each paragraph (Murata, 1999)

– In sentences that contain title words (Ko et al, 2002)

Sec. 15.3.2

Properties of Text

• Word frequencies - skewed distribution

• `The’ and `of’ account for 10% of all words

• Six most common words account for 40%

From [Croft, Metzler & Strohman 2010] 75

Associate Press Corpus `AP89’

From [Croft, Metzler & Strohman 2010] 76

Middle Ground

• Very common words bad features

• Language-based stop list:

words that bear little meaning

20-500 wordshttp://www.dcs.gla.ac.uk/idom/ir_resources/linguistic_utils/stop_words

• Subject-dependent stop lists

• Very rare words also bad features

Drop words appearing less than k times / corpus

77

Word Frequency

• Which word is more indicative of document similarity?

– ‘book,’ or ‘Rumplestiltskin’?

– Need to consider “document frequency”--- how frequently the

word appears in doc collection.

• Which doc is a better match for the query “Kangaroo”?

– One with a single mention of Kangaroos… or a doc that

mentions it 10 times?

– Need to consider “term frequency”--- how many times the

word appears in the current document.

78

TF x IDF

)/log(* kikik nNtfw

C in T term of frequency document inverse idf

D document in T term of frequencytf

D document in k termT

kk

ikik

ik

nNidf

kk log

kk T contain that C indocuments of number then

C collection the indocuments of number total N

79

Inverse Document Frequency

• IDF provides high values for rare words and

low values for common words

• Add 1 to avoid 0.

41

10000log

698.220

10000log

301.05000

10000log

010000

10000log

80

TF-IDF normalization

• Normalize the term weights

– so longer docs not given more weight (fairness)

– force all values to fall within a certain range: [0, 1]

t

k kik

kikik

nNtf

nNtfw

1

22 )]/log(1[)(

))/log(1(

81

Term-Weighted Complement Naïve Bayes

82

Multi-class Problems

•91

•92

• Most (over)used data set, 21,578 docs (each 90 types, 200 tokens)

• 9603 training, 3299 test articles (ModApte/Lewis split)

• 118 categories

– An article can be in more than one category

– Learn 118 binary category distinctions

• Average document (with at least one category) has 1.24 classes

• Only about 10 out of 118 categories are large

Common categories

(#train, #test)

Evaluation: Classic Reuters-21578 Data Set

• Earn (2877, 1087) • Acquisitions (1650, 179)• Money-fx (538, 179)• Grain (433, 149)• Crude (389, 189)

• Trade (369,119)• Interest (347, 131)• Ship (197, 89)• Wheat (212, 71)• Corn (182, 56)

Sec. 15.2.4

•93

Reuters Text Categorization data set (Reuters-21578) document

<REUTERS TOPICS="YES" LEWISSPLIT="TRAIN" CGISPLIT="TRAINING-SET" OLDID="12981"

NEWID="798">

<DATE> 2-MAR-1987 16:51:43.42</DATE>

<TOPICS><D>livestock</D><D>hog</D></TOPICS>

<TITLE>AMERICAN PORK CONGRESS KICKS OFF TOMORROW</TITLE>

<DATELINE> CHICAGO, March 2 - </DATELINE><BODY>The American Pork Congress kicks off tomorrow,

March 3, in Indianapolis with 160 of the nations pork producers from 44 member states determining industry positions

on a number of issues, according to the National Pork Producers Council, NPPC.

Delegates to the three day Congress will be considering 26 resolutions concerning various issues, including the future

direction of farm policy and the tax law as it applies to the agriculture sector. The delegates will also debate whether to

endorse concepts of a national PRV (pseudorabies virus) control and eradication program, the NPPC said.

A large trade show, in conjunction with the congress, will feature the latest in technology in all areas of the industry,

the NPPC added. Reuter

&#3;</BODY></TEXT></REUTERS>

Sec. 15.2.4

Precision & Recall

P

N

“P” “N”

TP FN

FP TN

Predicted

Act

ual

Two class situation

FP

FPTP

Multi-class situation:

Precision = TP/(TP+FP)

Recall = TP/(TP+FN)

F-measure = 2pr/(p+r)

94

Micro-‐ vs. Macro-‐Averaging

95

• If we have more than one class, how do we combine

multiple performance measures into one quantity?

• Macroaveraging

– Compute performance for each class, then average.

• Microaveraging

– Collect decisions for all classes, compute contingency table,

evaluate

Precision & Recall

Multi-class Multi-label situation:

96

Precision(class 1) = 251/(Column1)

Recall(class 1) = 251/(Row1)

F-measure(class 1)) = 2piri/(pi+ri)

Precision(class i) = TPi/(TPi+FPi)

Recall(class i) = TPi/(TPi+FNi)

F-measure(class i) = 2piri/(pi+ri)

Aggregate

Average Macro Precision = Σpi/N

Average Macro Recall = Σri/N

Average Macro F-measure = 2pMrM/(pM+rM)

Average Micro Precision = ΣTPi/ ΣiColi

Average Micro Recall = ΣTPi/ ΣiRowi

Average Micro F-measure = 2pμrμ /(pμ+rμ)

Precision & Recall

Multi-class situation:

97

Precision(class 1) = 251/(Column1)

Recall(class 1) = 251/(Row1)

F-measure(class 1)) = 2piri/(pi+ri)

Precision(class i) = TPi/(TPi+FPi)

Recall(class i) = TPi/(TPi+FNi)

F-measure(class i) = 2piri/(pi+ri)

Aggregate

Average Macro Precision = Σpi/N

Average Macro Recall = Σri/N

Average Macro F-measure = 2pMrM/(pM+rM)

Average Micro Precision = ΣTPi/ ΣiColi

Average Micro Recall = ΣTPi/ ΣiRowi

Average Micro F-measure = 2pμrμ /(pμ+rμ)

Aren’t μ prec and μ recall the same?

Missed predictions

Classifier hallucinations

Multi-Class Classification

• What?

– Converting a k-class problem to a binary problem.

• Why?

– For some ML algorithms, a direct extension to the multiclass case may be problematic.

• How?

– Many methods

•99

Methods

• One-vs-all

• All-pairs

• Error-correcting Output Codes (ECOC)

• …

•100

One-vs-all

• Idea:

• Create many 1 vs other classifiers

– Classes = City, County, Country

– Classifier 1 = {City} {County, Country}

– Classifier 2 = {County} {City, Country}

– Classifier 3 = {Country} {City, County}

• Training time:

– For each class cm, train a classifier clm(x)

• replace (x,y) with

(x, 1) if y = cm

(x, -1) if y != cm•101

An example: training

• x1 c1 …

• x2 c2 …

• x3 c1 …

• x4 c3 …

for c1-vs-all:

x1 1 …

x2 -1 …

x3 1 …

x4 -1 …

for c2-vs-all:

x1 -1

x2 1 …

x3 -1 …

x4 -1 …

for c3-vs-all:

x1 -1…

x2 -1…

x3 -1 …

x4 1 …

•102

One-vs-all (cont)

• Testing time: given a new example x

– Run each of the k classifiers on x

– Choose the class cm with the highest

confidence score clm(x):

c* = arg maxm clm(x)

•103

An example: testing

• x1 c1 …

• x2 c2 …

• x3 c1 …

• x4 c3 …

three classifiers

Test data:

x ?? f1 v1 …

for c1-vs-all:

x ?? 1 0.7 -1 0.3

for c2-vs-all

x ?? 1 0.2 -1 0.8

for c3-vs-all

x ?? 1 0.6 -1 0.4

=> what’s the system prediction for x?

•104

All-pairs (All-vs-All (AVA))

• Idea:

– For each pair of classes build a classifier

– {City vs. County}, {City vs Country}, {County vs.

Country}

– Ck2 classifiers: one classifier for each class pair.

• Training:

– For each pair (cm, cn) of classes, train a classifier clmn

• replace a training instance (x,y) with

(x, 1) if y = cm

(x, -1) if y = cn

otherwise ignore the instance

•105

An example: training

• x1 c1 …

• x2 c2 …

• x3 c1 …

• x4 c3 …

for c1-vs-c2:

x1 1 …

x2 -1 …

x3 1 …

for c2-vs-c3:

x2 1 …

x4 -1 …

for c1-vs-c3:

x1 1…

x3 1 …

x4 -1 …

•106

All-pairs (cont)

• Testing time: given a new example x

– Run each of the Ck2 classifiers on x

– Max-win strategy: Choose the class cm that wins the most pairwise comparisons:

– Other coupling models have been proposed: e.g., (Hastie and Tibshirani, 1998)

•107

An example: testing

• x1 c1 …

• x2 c2 …

• x3 c1 …

• x4 c3 …

three classifiers

Test data:

x ?? f1 v1 …

for c1-vs-c2:

x ?? 1 0.7 -1 0.3

for c2-vs-c3

x ?? 1 0.2 -1 0.8

for c1-vs-c3

x ?? 1 0.6 -1 0.4

=> what’s the system prediction for x?

•108

Hierarchical Categorization

• Pick the category with max probability

• Create many OVA/AVA classifiers

• Use a hierarchical approach (wherever

hierarchy available)

Entity

Person Location

Scientist Artist City County Country

109

Error-correcting output codes

(ECOC)

• Proposed by (Dietterich and Bakiri, 1995)

• Idea:

– Each class is assigned a unique binary string of length n.

– Train n classifiers, one for each bit.

– Testing time: run n classifiers on x to get a n-bit string s, and choose the class which is closest to s.

•110

An example: Digit Recognition

•111

Meaning of each column

•112

Another example: 15-bit code for a

10-class problem

•113

Hamming distance

• Definition: the Hamming distance between two strings of equal length is the number of positions for which the corresponding symbols are different.

• Ex:

– 10111 and 10010

– 2143 and 2233

– Toned and roses

•114

How to choose a good error-

correcting code?

• Choose the one with large minimum

Hamming distance between any pair of

code words.

• If the min Hamming distance is d, then the

code can correct at least (d-1)/2 single bit

errors.

•115

Two properties of a good ECOC

• Row separations: Each codeword should be

well-separated in Hamming distance from

each of the other codewords

• Column separation: Each bit-position

function fi should be uncorrelated with each

of the other fj.

•116

Summary

• Different methods:

– Direct multiclass, if possible

– One-vs-all (a.k.a. one-per-class): k-classifiers

– All-pairs: Ck2 classifiers

– ECOC: n classifiers (n is the num of columns)

– Hierarchical classification (logC classifiers)

• Some studies report that All-pairs and

ECOC work better than one-vs-all.

•120